cross-encoder/nli-deberta-v3-xsmall: open-weights language model for zero-shot classification | Hugging Bay

Skip to main contentHugging Bay

cross-encoder/nli-deberta-v3-xsmall

cross-encoder/nli-deberta-v3-xsmall is an open-weights language model for zero-shot classification under the apache-2.0 license (2.1 GB), released Apr 11, 2025. Hosted open-weights model — 7 files carry published SHA-256 checksums you can re-hash. 7 Hugging Ba

License: apache-2.0
Task: zero-shot classification
Size: 2.1 GB
Recommended file: 270.2 MB — model.safetensors
Files: 7 hosted of 7 listed
Download: Download here (7 hosted files)
Released: Apr 11, 2025
Updated on Hugging Bay: Jul 31, 2026
Source: cross-encoder/nli-deberta-v3-xsmall

FAQ

What is cross-encoder/nli-deberta-v3-xsmall?

cross-encoder/nli-deberta-v3-xsmall is an open-weights language model for zero-shot classification published by cross-encoder, released Apr 11, 2025. Model from Hugging Face for zero shot classification: cross-encoder/nli-deberta-v3-xsmall · sentence transformers

What license does cross-encoder/nli-deberta-v3-xsmall use, and can I use it commercially?

It is released under the apache-2.0 license. Commercial use is allowed. Permissive open license: commercial use, modification, and redistribution are permitted.

How big is cross-encoder/nli-deberta-v3-xsmall?

The full upstream repository is 2.1 GB; the recommended download, model.safetensors, is 270.2 MB.

Where can I download cross-encoder/nli-deberta-v3-xsmall?

Hugging Bay hosts 7 of 7 files for this model, each with a published SHA-256 checksum. Total size: 2.1 GB. The original release is at https://huggingface.co/cross-encoder/nli-deberta-v3-xsmall.

Has cross-encoder/nli-deberta-v3-xsmall been scanned for unsafe files?

Source scan status: passed. 7 hosted files carry a published SHA-256 you can re-hash. The machine-readable verdict is at /api/artifacts/hf-model-cross-encoder-nli-deberta-v3-xsmall/safety on Hugging Bay.

Does cross-encoder/nli-deberta-v3-xsmall run on an RTX 4090 (24GB)?

cross-encoder/nli-deberta-v3-xsmall needs ~2.16 GB, which is within the 24 GB on an RTX 4090 (24GB) by memory capacity (~21.84 GB headroom). Computed from stated size and quant, not a runtime guarantee.

Does cross-encoder/nli-deberta-v3-xsmall run on an RTX 3090 (24GB)?

cross-encoder/nli-deberta-v3-xsmall needs ~2.16 GB, which is within the 24 GB on an RTX 3090 (24GB) by memory capacity (~21.84 GB headroom). Computed from stated size and quant, not a runtime guarantee.

Does cross-encoder/nli-deberta-v3-xsmall run on an A100 (80GB)?

cross-encoder/nli-deberta-v3-xsmall needs ~2.16 GB, which is within the 80 GB on an A100 (80GB) by memory capacity (~77.84 GB headroom). Computed from stated size and quant, not a runtime guarantee.

Does cross-encoder/nli-deberta-v3-xsmall run on a 16GB laptop (no dedicated GPU)?

cross-encoder/nli-deberta-v3-xsmall needs ~2.16 GB, which is within the 16 GB on a 16GB laptop (no dedicated GPU) by memory capacity (~13.84 GB headroom). Computed from stated size and quant, not a runtime guarantee.

Does cross-encoder/nli-deberta-v3-xsmall run on an 8GB laptop (no dedicated GPU)?

cross-encoder/nli-deberta-v3-xsmall needs ~2.16 GB, which is within the 8 GB on an 8GB laptop (no dedicated GPU) by memory capacity (~5.84 GB headroom). Computed from stated size and quant, not a runtime guarantee.

About this artifact

External Hugging Face catalog record for cross-encoder/nli-deberta-v3-xsmall. Task: zero shot classification. Library: sentence transformers. Languages: en. Datasets referenced by card metadata: nyu-mll/multi_nli, stanfordnlp/snli. Popularity signal at import: 163,249 downloads and 7 likes. Files are listed for discovery and verification context; Hugging Bay serves local downloads only after explicit mirror review, scanning, hashing, and signing.

Hosted file inventory

File inventory only. Individually hashed hosted files may be available, but bundle completeness, verification, and runtime state are unknown.

Hosted files: 7 of 7
Reviewed peer-assisted fallbacks: 0
Trust bundle: Trust-bundle JSON
Review summary: Review-summary JSON
Distribution: Reviewed peer-fallback JSON

Run status

Not hosted here yet — Not hosted here yet — the mirror is in progress. Check back or request it.

Detailed evidence for the selected file set

This detail applies to the selected file set, not every file in the repository.

Executable state: complete-bytes
Evidence timestamp: 2026-07-31T01:34:14.832Z
Selected representation: rep-de320b98dbbe57af4c79e3edef08a9794bd95fbf83885d63334f97523d55cda9
Complete bytes: yes
Complete kit: no
Verification: not verified
Run ready: not established
Files: View file metadata

Bay Run

Try It Now

Run Bay Run's Support-ticket Pin: bounded zero-shot routing over exactly three labels (access / billing / delivery). Provisional and unmeasured.

Call Bay Run's run_pin MCP tool at https://run.huggingbay.xyz/mcp/ with Pin route_571826c40685073a99510b1951e60338, or POST https://run.huggingbay.xyz/v1/run/route_571826c40685073a99510b1951e60338. Served model: cross-encoder/nli-deberta-v3-xsmall · labels access_issue, billing_issue, delivery_issue.

run_pin arguments

{
  "pin_id": "route_571826c40685073a99510b1951e60338",
  "input": "I cannot log into my account."
}

Copy agent call

# 1. mint a $0 demo bearer (no signup)
TOKEN=$(curl -fsS -X POST https://run.huggingbay.xyz/oauth/token -H 'Content-Type: application/json' \
  -d '{"grant_type":"urn:bay-run:grant-type:demo","scope":"mcp:demo"}' | python3 -c 'import sys,json;print(json.load(sys.stdin)["access_token"])')
# 2. run the Pin (MCP run_pin) — the same next_call find_runnable and the search row return
curl -fsS -X POST https://run.huggingbay.xyz/mcp/ -H "Authorization: Bearer $TOKEN" \
  -H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"run_pin","arguments":{"pin_id":"route_571826c40685073a99510b1951e60338","input":"I cannot log into my account."}}}'
# read decision.action first; the receipt in structuredContent is the proof of execution

Exact live run contract JSON

Request a complete representation

License: pass - apache-2.0
Provenance: pass - cross-encoder/nli-deberta-v3-xsmall
Scan: pass - passed
Hosted files: warn - 7/7 repository files hosted; selected representation evaluated separately
Byte manifest integrity: warn - not final yet

Machine-readable card

Open interactive artifact page

Evidence details

Demand: 187K upstream downloads
Trust: Metadata checked 0/100
Trusted download: File inventory only
Run fit: Canonical selected-byte evidence recorded; public obtain path unavailable
Scan status: passed
Executable state: complete-bytes
Evidence timestamp: 2026-07-31T01:34:14.832Z
Declared license: apache-2.0
Pipeline: zero-shot-classification
Library: sentence-transformers
Task categories: zero-shot-classification
Languages: en
Referenced datasets: nyu-mll/multi_nli, stanfordnlp/snli
Metrics: accuracy
File hints: 16 upstream file samples

Public files only. Gated and restricted sources stay gated.